Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Codec Killers (Read 10872 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Codec Killers

Reply #1 –
Its not a bad article really. Good points from monty about potential pitfalls of lossy transcoding as well as crappy P2P quality. They also mention than lossy cannot be transparent all the time - esp at lower bitrates. This might make some think twice about buying lower quality lossy files.

One problem I see is that a lot of these 'off' feelings might be pure imagination unless you have the original to abx and those reading this will become more paranoid than before.

Codec Killers

Reply #2 –
I think this is the first time I have seen any discussion in the "mainstream" news about the possiblity that the music being purchased online, may not be as good as the CD purchased in the store.
flac > schiit modi > schiit magni > hd650

Codec Killers

Reply #3 –
I think it's nice to have some discussion on how lossy encoding works showing up in places more mainstream than HydrogenAudio, but I'm a little miffed at how the article doesn't mention lossless encoding as an alternative.

Codec Killers

Reply #4 –
unfortunately the majority of people simply don't care :/
Angels fall farthest...

Codec Killers

Reply #5 –
Quote
unfortunately the majority of people simply don't care :/
[a href="index.php?act=findpost&pid=249337"][{POST_SNAPBACK}][/a]


Very true.  We must remember that for most people the sound quality is not very important (loudness race, poorly encoded P2P files, etc.).  I suspect DRM issues and compatability are much more important to the average consumer.

I don't think a couple of glitchy spots in a few songs is really that important.  Even some artifacting will be preferable to a scratched/unplayable CD which many people tolerate now.

>>Edit: artifact comment<<
Was that a 1 or a 0?

Codec Killers

Reply #6 –
I've said this before; extreme compression and limiting of music prior to CD mastering (almost mandatory nowadays with 'popular' titles) makes the content of these discs much more amenable to lossy digital compression. They are effectively using, say,  12-bits or less of resolution.

Perhaps the 'loudness war' is indeed simply the result of industry philistinsim, but it's now coincidentally proving very useful for touting so-called "CD-quality" 128-KbS AAC's WMA's etc.

Hmmm.

RF.

Codec Killers

Reply #7 –
In the article, it states;

"....Songs that are recorded at very high volumes can prove difficult,....."

Song' with very high *swings* in volume, using a lot of dynamic range are difficult. A pop CD that relentlessly uses only the top 6dB of 16/44 is surely the easiest for most lossy codecs?

RF

Codec Killers

Reply #8 –
Quote
In the article, it states;

"....Songs that are recorded at very high volumes can prove difficult,....."

Song' with very high *swings* in volume, using a lot of dynamic range are difficult. A pop CD that relentlessly uses only the top 6dB of 16/44 is surely the easiest for most lossy codecs?

RF
[a href="index.php?act=findpost&pid=249497"][{POST_SNAPBACK}][/a]


I don't know how lossless compressing ratios correlate with lossy ratios (if at all), but my worst losslessly compressing album also happens to be the loudest and with the most squashed dynamics. Classical music with a low average level (but very dynamic music otherwise) tends to compress with excellent ratios.

Codec Killers

Reply #9 –
i find it funny the MS guy is describing pre-echo.  he must be VERY familiar with it by now 

good that they gave vorbis a mention.

oh, and loud songs tend to compress worse because they have added high frequencies which lead to bloat (well, that's my guess).  loud guitar stuff usually isn't affected as much though, nor really noise-like stuff.

Codec Killers

Reply #10 –
A common problem with high volume recordings is the clipping, usually inaudible in the original, that gets worse without using replaygain. The added clipping during the playback is often audible. It can sound like clicks, but also like ringing.

Some original compressed recordings features this clipping, and, listening to them, I always wonder if my musepack file rings, but no, most of the times, it's the original that is clipped.

Codec Killers

Reply #11 –
Quote
In the article, it states;

"....Songs that are recorded at very high volumes can prove difficult,....."

Song' with very high *swings* in volume, using a lot of dynamic range are difficult. A pop CD that relentlessly uses only the top 6dB of 16/44 is surely the easiest for most lossy codecs?

RF
[a href="index.php?act=findpost&pid=249497"][{POST_SNAPBACK}][/a]


I'm not an expert in perceptual coding, but I think the opposite is true.  If a song is hugely compressed and loud, it's not using only the top 6 db, it's using up the entire dynamic range of the medium with every single peak and valley.

A recording that has good dynamics will only have a few peaks per song that reach digital full scale and use up the entire 16 bits.

Therefore the song that reaches full scale on every note contains more information than the song with good dynamics, even though it surely sounds far worse.  This is why hypercompressed songs tend to have very high bitrates when encoded to lossy or lossless, compared to a track with lower volume.

PS: RockFan, I still think your theory about awful modern mastering and "cd quality" claims being connected is a good one, except it's due to the fact that the music already sounds so awful that virtually no lossy artifacts could make it much worse. 

Codec Killers

Reply #12 –
True. Look at the current 128k tests and most are either struggling to find artifacts or don't find them to be that annoying. Take the modern compressed cd and the faults are much easier to identify and far more annoying than 128k artifacts (to me anyway).

I think mp3 bad quality is exaggerated, due to old blade, xing encoders and transcoding. People should start focusing on how crappy cd music sounds nowdays.

Codec Killers

Reply #13 –
Quote
A common problem with high volume recordings is the clipping, usually inaudible in the original, that gets worse without using replaygain. The added clipping during the playback is often audible. It can sound like clicks, but also like ringing.
[a href="index.php?act=findpost&pid=249581"][{POST_SNAPBACK}][/a]

I agree.

Recently, the MSN music service became available here in Belgium, and just out of curiosity I decided to buy a few tracks.

The clipping was awful.

And since the files were DRM'ed WMA files, I had to use Windows Media Player for playback, so replaygain was out of the question.

Unfortunately, I think nowadays the kids are so used to clipping that they don't hear it anymore.

OT: Since I mentioned the MSN music service, I just want to vent a little: when are we finally going to see a good unlimited streaming service in Belgium (like e.g. Rhapsody, Napster or MusicMatch MX)? I've been waiting for years already.
Over thinking, over analyzing separates the body from the mind.

Codec Killers

Reply #14 –
Am I right in saying that there's a difference between clipping and hard-limiting?

The former seems to  the result of simply driving a digital recording into 'overs'. the latter using an algorithm (or an analogue circuit) that attempts to modify the peaks slightly to a downward sloping shape as they hit, erm, a hard limit.

I get the feeling that using digital 'overs' is a quick-and-dirty (read; lazy) way of accomplishing high RMS values (not that hard-limiting is desirable in the first place).

RF.

Codec Killers

Reply #15 –
Quote
A recording that has good dynamics will only have a few peaks per song that reach digital full scale and use up the entire 16 bits.

Therefore the song that reaches full scale on every note contains more information than the song with good dynamics, even though it surely sounds far worse.  This is why hypercompressed songs tend to have very high bitrates when encoded to lossy or lossless, compared to a track with lower volume.

PS: RockFan, I still think your theory about awful modern mastering and "cd quality" claims being connected is a good one, except it's due to the fact that the music already sounds so awful that virtually no lossy artifacts could make it much worse. 
[a href="index.php?act=findpost&pid=249594"][{POST_SNAPBACK}][/a]


A while ago I (re)acquired the entire Matt Johnson (The The) back catalogue, re-released in 2002.

"24 bit remastered and restored, to reveal the richness and complexity of the original recordings" it says on the swanky slip-sleeved packaging.

I'm not quite sure why they needed 'restoring', but they were certainly loud, and frankly didn't sound like I remembered them (on vinyl, admittedly).

So over a couple of weeks I obtained the original late-80's early 90's CD versions. Much better  - (IMO, although the older  'Soul Mining' was a bit 'gritty' sounding) - those remasters had indeed been heavily compressed, to make them sound more 'modern'.

On a track like 'Violence Of Truth' on Mind Bomb, A/B'ing the 2 versions (volume up on the original to match the levels) revealed that the startlingly hard-edged snare-drum had largely lost it's impact on the new version. It had already been peaking at or near full-scale, so as everything else had got louder under it, it had 'nowhere to go' when the compressor had been applied.

It would be interesting to use such wildely different versions of otherwise identical tracks to test lossy codecs - it's a pity I disposed of the remasters!

ciao,
RF

Codec Killers

Reply #16 –
There real issue is that most consumers don't know because because they aren't really listening.  Adn since they don't know - they don't care.  Most the garbage being sold is being played while jogging, reading, studying, driving, etc.  It is background noise only.  Most buyers have grown-up to only listening to the radio or only listening on their $49 CD player with their $12 headphones.  They hear no difference between a 128 Xing MP3 they found on Kazaa, the radio, and their favorite CD. 

Indeed, most music sold is not even intended to be listened to with any effort.  It is the fast food version of sound meant to be consumed quickly, on the go, with minimum effort.  These stores will only get bigger as the next generation of listeners gets used to the even lower quality of sound these places sell.  I am suprised that no one has yet tried to sell a "best of" CD compiliation of multible albums on one CD in mp3 format.  99% of the buyers could tell no difference and looky, looky - 100 songs on a CD.  And it sounds even better tham the old fashioned seperate CDs!!!

Codec Killers

Reply #17 –
Quote
Its not a bad article really. Good points from monty about potential pitfalls of lossy transcoding as well as crappy P2P quality. They also mention than lossy cannot be transparent all the time - esp at lower bitrates. This might make some think twice about buying lower quality lossy files.

One problem I see is that a lot of these 'off' feelings might be pure imagination unless you have the original to abx and those reading this will become more paranoid than before.
[a href="index.php?act=findpost&pid=249314"][{POST_SNAPBACK}][/a]


I think that's a resonably sensible article - I'm impressed. One thing that does get up my nose is reading repeated articles from so-called "technology" journo's who assert how wma is the best format for portable music without presenting any facts or data to support it.

Codec Killers

Reply #18 –
Quote
[...]digital music companies[...]must ultimately convince online music buyers that computer files provide a quality listening experience.

I have never bought digital songs from Apple or anywhere else, but when people buy digital songs from those places, do the files come with some kind of disclaimer saying that the music is lower in quality than CDs or words to that effect?

If not, they should at least compile a list of songs that get compressed badly (with ovbious artifacts), and work on those songs with a higher bitrate and offer that higer bitrate version to consumers by default. Pretending to be unaware of defects (artifacts) in their products (digital music) is unethical to say the least.

ADDITION (afterthought): I very much doubt corporations would bother with ideas like mine above.

Codec Killers

Reply #19 –
Quote
I am suprised that no one has yet tried to sell a "best of" CD compiliation of multible albums on one CD in mp3 format.  99% of the buyers could tell no difference and looky, looky - 100 songs on a CD.  And it sounds even better tham the old fashioned seperate CDs!!!
[a href="index.php?act=findpost&pid=249712"][{POST_SNAPBACK}][/a]



Ah, but they do - I have seen east Eurpoean bootleggers flogging such items on eBay, luckily with no success.

Codec Killers

Reply #20 –
Quote
I am suprised that no one has yet tried to sell a "best of" CD compiliation of multible albums on one CD in mp3 format.  99% of the buyers could tell no difference and looky, looky - 100 songs on a CD.  And it sounds even better tham the old fashioned seperate CDs!!!


100 songs on 700 MB sounds a bit like quite “high” bitrates to sell to people. I get 100-120 songs on my –alt-preset standard-powered daily mp3cdrws for my mp3cdplayer. Some of my friends who are “128 kbps music store files are good enough for me”-types  get 150+ files on their CD:s, so if someone would sell mp3cd:s encoded with –alt-preset standard I could buy it if it was a bit cheaper than buying it all as CD-Audio, because –alt-preset standard is what I listen too most of the time anyway.  But I guess some people would see the files as too large and transcoding to lower rates for portable devices as too complicated.

That point earlier in this thread about record companies offering nondynamic music so that it would sound better with lossy codecs was interesting, even if it’s proven wrong (?). I’m no expert, but my guess is that when record companies realize that even more of their most popular (=worst) music is being sold online (this should happen soon, if it hasn’t happened already) they will test the songs with different encoders in the mastering process.

To evolve my highly non-professional guess; I believe that if the popularity of online stores and lossy codec players continue to increase like it has done until now, mastering/studio software will sooner or later include “codec optimization” tools of some kind. Even if the encoders are evolved and improved all the time, I doubt that 128 kbps of any lossy codec will be what most people here would agree to buy their music in within a close future.

I had an interesting music class in school last fall. It was about producing some kind of song with Logic 6 on a Mac and when I had my song nearly done and I was polishing the panning, eq and other mastering related aspects I noticed one thing. I had quite a lot of different test versions of the song, that I carried around in my mp3-cd player. When I coded my daily test versions of the song using lame –alt-preset standard bitrate I noticed how much I was told about the complexity of my music just by watching the average bitrate –alt-preset standard produced. Certain instruments added a lot of aps bitrate if they weren’t placed in the middle of the sound field and when I played around with very aggressive stereo usage (imitating a “The Doors” or “Motown” kind of feeling) with lots of reverbs and ambience I realized how interesting it would be to do it the other way… Like getting the song to sound as complex as possible without doing it all too difficult for Joint Stereo algorithms and avoiding a need for a higher bitrate in VBR modes, you see?

I dunno anything about how psychoacoustic tricks work, but I’m sure that it can be involved in the mastering process to enable reverbs and panning without stressing encoders on low bitrate music store modes. I'm quite sure that DTS were doing that until the point when they startied to sell their encoder instead of encoding every DTS movie soundtrack "in house" as they did until 1999.

Codec Killers

Reply #21 –
The mentioned "Microsoft audio architect Jim Johnston" is James D. Johnston, aka JJ, one of the fathers of perceptual compresion, and of AAC and MP3 formats. People involved in audio compression development know well about him. He's one of the most knowledgeable people in audio I've read from at Internet. He worked at AT&T labs when he was involved in AAC development. He retired, and then went to Microsoft. I guess he's the main reason why WMA Pro is so much better than WMA Standard.

At the article they also asked Monty. It's not strange it contains accurate information.

Codec Killers

Reply #22 –
I assume it is the same J.D. Johnston who has a set of subband filters named after him

J.D. Johnston, ``A filter family designed for use in quadrature mirror filters banks', in Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing, 1980, pp. 291-294.

Codec Killers

Reply #23 –
Quote
I assume it is the same J.D. Johnston who has a set of subband filters named after him

J.D. Johnston, ``A filter family designed for use in quadrature mirror filters banks', in Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing, 1980, pp. 291-294.
[a href="index.php?act=findpost&pid=263971"][{POST_SNAPBACK}][/a]


I guess so ... it's the only thing I've in mind - the Johston-QMF filters. Anything else he came up with and I'm not aware of ?

SebastianG